Individual labour market transitions of Australians during and after the National COVID-19 Lockdown

ABSTRACT

We examine the individual labour market transitions of Australians during and after the National COVID-19 Lockdown, controlling for demographic characteristics and person fixed effects across different subgroups of the population using the Longitudinal Labour Force Survey. The National COVID-19 Lockdown (which began on 21 March 2020 with the introduction of social distancing rules and the closure of non-essential services across individual states and territories and lasted until the end of June 2020) decreased the overall labour force participation by 3% and increased unemployment by 1.8%. However, the economy recovered to a certain extent after the lockdown, with labour force participation increasing by 0.051% and unemployment declining by 0.049% for each additional week after the end of the lockdown. Our conditional estimates show that the national lockdown did not affect the genders differently in terms of unemployment, while females recovered faster during the post-lockdown period. People working in transport, postal, administrative, and arts and recreation services decreased their working hours significantly during the lockdown relative to those employed in other industries, but we do not observe any significant difference in their post-lockdown recovery patterns. Our results could help policy makers better target the labour market outcomes of the most at-risk individuals.

KEYWORDS:

• Australia
• National Lockdown
• Personal Labour Market Behaviour
• COVID-19

JEL CLASSIFICATION:

• I15
• I18
• J21

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00036846.2022.2094881

Notes

1. The national lockdown was implemented as follows. A person could leave his/her house for only four reasons (grocery shopping, medical care, daily exercise and going to work), no one could have visitors at home, and everyone must maintain a distance of 1.5 metres from others in public places. Moreover, all non-essential services were shut down because of the social distancing rules. Pubs, bars and nightclubs, as well as all entertainment and cultural venues, were closed, while restaurants, cafes and bottle shops were take-away only. Beauty services were also closed, but hairdressers remained open. Only shopping centres, markets and other retail shops selling essentials were still allowed to trade, and these were subject to the social distancing rules.

2. For example, the Victorian State Government implemented a second (state-specific) lockdown from July to November 2020 due to an outbreak associated with hotel quarantine. However, as we will explain further later, we focus our analysis on the labour market impacts of the overall national lockdown, which covered all states.

3. Eligible businesses that had suffered a significant loss (about 30%–50% of GST turnover) were entitled to a fortnightly payment of 1500 AUD for each eligible employee.

4. To be eligible to receive the JobSeeker payment, a person must be between 22 and 70, an Australian permanent resident or citizen, unemployed, and either looking for a job or sick and unable to work.

5. We calculated the effect at the mean level by dividing the coefficient by the mean value of the dependent variable. For example, we show that the national lockdown led to a 36% decrease at the mean level in the unemployment, which we calculate by dividing the estimated coefficient (0.0180) by the mean value of the unemployment (0.049) in our data.

6. In addition, Beland et al. (2020b) report that COVID-19 led to a 5% increase in unemployment and a 3.7% decrease in LFP in Canada.

7. Given the definitions of the two balanced panel samples, it is clear that they include two entirely different sets of people, as there is no overlap of respondents. We present the summary statistics for both samples in Appendix Tables 1 and 2.

8. The authors classify all occupations by giving them a value from 0 to 1 depending on whether a job can be done from home. For example, they argue that computer and mathematical occupations can be done entirely from home and therefore assign them a value of 1, while building, ground cleaning and maintenance occupations cannot be done from home and are therefore assigned a value of 0. The remaining occupations are classified between those two bounds. We construct a Work from Home index for our samples using the same methodology and standardise it to have a mean of 0 and standard deviation of 1, to facilitate interpretation.

9. Unreported regressions reveal that our estimates are robust to over-controlling. More specifically, we include the controls and fixed effects used for our analysis one-by-one and show that our estimates are robust to the inclusion of such a rich set of covariates. We also replaced the state of residence fixed effects with labour market region fixed effects and the estimates remained significant and similar in magnitude. The results are available upon request.

10. Appendix Table 3 supplements the binary indicator results and checks their robustness to our underlying assumption of a one-time effect by presenting a categorical variable, Weeks in National Lockdown, which measures the number of weeks that have passed since the lockdown was implemented in different categorical groups. This variable was set equal to 0 for the weeks before the lockdown was enacted, as the reference or excluded category. Our results show that the effects of the national lockdown on LFP and unemployment remain similar in magnitude across all categories. In addition, we find that the adverse impact of the lockdown on full-time employment and working hours was stronger during the first four weeks of the lockdown, while the effect on the probability of having a single job compared to having multiple jobs comes from the first six weeks of the lockdown. Lastly, we find that the propensity for self-employment (vs. employee) increased during the last couple of weeks of the lockdown. Overall, these results seem to be consistent with the results of the baseline specification.

11. Appendix Table 4 also examines potential non-linearities in the effects of the reversal of the national lockdown on labour market outcomes. We generate a categorical variable that measures the number of weeks since the national lockdown ended. This enables us to examine whether the effect of each additional week post-lockdown on the labour market changes over time. Our estimates are in line with our baseline specification, showing that the effect becomes stronger as the number of weeks since the lockdown ended increases.